In the Hadoop context, which processing approach breaks a task into smaller elements and processes them on multiple servers?

• A. Data mining
• B. MapReduce
• C. In-memory processing
• D. Data replication

Answer: (B)

The main concept being tested is distributed parallel processing across a cluster using a model that splits work into smaller pieces. In Hadoop, MapReduce breaks a large task into many map tasks that run in parallel on different servers, each handling a chunk of data and producing intermediate results. Those results are then shuffled and fed to reduce tasks, which combine them to form the final output. This approach enables scalable processing, fault tolerance, and close data locality by performing computation where the data resides. The other terms describe different ideas: data mining is about discovering patterns in data, in-memory processing speeds operations by keeping data in RAM, and data replication focuses on duplicating data for reliability rather than how tasks are executed across servers.